pcp
[Top] [All Lists]

Re: pmie spawning more than 1 instance per host

To: Jan-Frode Myklebust <janfrode@xxxxxxxxx>
Subject: Re: pmie spawning more than 1 instance per host
From: Nathan Scott <nscott@xxxxxxxxxx>
Date: Wed, 13 Feb 2008 11:24:04 +1100
Cc: pcp@xxxxxxxxxxx
In-reply-to: <slrnf5nl5l.1lc.mykleb@xxxxxxxxxxxxxxxxxxxxxxx>
Organization: Aconex
References: <slrnf5nl5l.1lc.mykleb@xxxxxxxxxxxxxxxxxxxxxxx>
Reply-to: nscott@xxxxxxxxxx
Sender: pcp-bounce@xxxxxxxxxxx
On Tue, 2007-05-29 at 09:22 +0200, Jan-Frode Myklebust wrote:
> I'm using pmie to monitor about 50 hosts from a central monitor. The
> central monitor is running the pmie_check from /etc/cron.hourly/, 
> and annoyingly it seems to not always be able to detect if an instance
> for a host is already running, so after a few days, I end up with more
> than one pmie per host.
> 
> Anyone else seen this? And maybe have a workaround?
> 
> Running v2.7.1 on RHEL4 as the monitoring host, but saw the same 
> problem on v2.5.0. Clients are a mix of mainly v2.5.0 and v2.7.1. All
> RHEL4/RHEL5.
> 

Hi Jan-Frode,

We recently came across this exact problem in our production monitoring
as well.  I narrowed ours down to a spurious DNS hostname aliasing issue
that was causing mismatching on hostname in pmie_check.sh.  I've put the
fix into my pcp git tree, you can pull it out of there if you like:
http://oss.sgi.com/cgi-bin/gitweb.cgi?p=nathans/kmchart.git;a=shortlog
If you're still using pmie, please let me know if it resolves the issue
for you too - thanks!

cheers.

-- 
Nathan


<Prev in Thread] Current Thread [Next in Thread>